Quaternion Polar Complex Exponential Transform and Local Binary Pattern-Based Fusion Features for Content-Based Image Retrieval

The presented work is a sincere effort to exhibit content-based image retrieval (CBIR) system utilized for extracting images from large databases. It makes use of image features like texture and color. Local binary pattern (LBP)-based operator brings about the information related to image texture by taking into consideration the surrounding pixel values. In spite of having its own advantages, this feature is not that superior at capturing the image’s information related to color. The present work rectifies this disadvantage by incorporating an extra color feature by the name quaternion polar complex exponential transform (QPCET) in addition to the LBP-based feature in the image retrieval system. The integrated QPCET- and LBP-based CBIR system exhibits better average retrieval efficiency compared to other accessible techniques tested on different benchmark databases.